[Bug 1899308] Re: failure to boot groovy daily (again)

Thomas Schmitt 1899308 at bugs.launchpad.net
Tue Oct 13 10:44:58 UTC 2020


Hi,

sudodus wrote:
> I also tested the tweak in comment #15 on the current Ubuntu Groovy iso file.
> When cloned the USB drive fails to boot the Lenovo V130,
> ...
> $ sudo gdisk test_mbr.iso
>   MBR: MBR only

Obviously the production of "test_mbr.iso" did what the file name says
and made a MBR partition table. Your reported run in comment #15 is
supposed to produce GPT for the Lenovos.

> Anyway, the tweak in comment #21 failed for me.

That one is aimed to produce an MBR table in order to let Chris Guiver
test whether this works with his HPs.
We know that this does not work for the Lenovos.

> maybe because something is 'set twice'.

Hardly. If settings in xorriso are in conflict, the later command overrides
the setting of the earlier command.
E.g. in the comment #15 with groovy-desktop-amd64_20201010_gpt.iso as base
  -boot_image any replay
sets appended_part_as=mbr, but the later command
  -boot_image any appended_part_as=gpt
overrides this.
When the production of the ISO filesystem begins, the setting is "gpt".

> Please notice that the option -compliance no_emul_toc may cause
problems.

Do you have any tangible indications for this (beyond the fact that MBR
table does not work for the Lenovos) ?

> Would it be useful for you, if I upload the modified iso file

It would suffice to show the exact xorriso run arguments, the URL of
the original ISO, and the inspecting program runs of xorriso or partition
editors with their results.

-------------------------------------------------------------------------

Whatever, we need to know whether Chris' hp dc7700 and hp dc7700 boot
by an MBR partition table. See comment #21.

Further we need confirmation that they still not boot with the newest
daily groovy ISO. (The one with GPT which has the boot flag attributed to
a nearly non-existing second MBR partition.)

If it turns out that the HPs only boot via MBR table, and if Ubuntu still
wants to have its ISO booting on those machines and on Lenovos, then i
fear that two ISO variations are needed:

- For most machines: GPT without that fabricated boot flag, which is a
  hack that is barely UEFI specs compliant.

- For the HPs and maybe other old machines: MBR partition table.
  (That variant has a boot flag by default.)

Most machines will work with both ISOs.

As said, Debian does similar for machines which are allergic to the normal
Debian EFI boot equipment. (The "mac" netboot ISOs are simply BIOS-only.
No special Apple parts inside.)

If only a single ISO is possible, then it should stay as it is now (GPT and
boot flag hack) which tosses some of the HPs out of the game.
(We now know how a user can repack that ISO to a HP-friendly one.)

Have a nice day :)

Thomas

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to cd-boot-images-amd64 in Ubuntu.
https://bugs.launchpad.net/bugs/1899308

Title:
  failure to boot groovy daily  (again)

Status in Ubuntu CD Images:
  Fix Released
Status in cd-boot-images-amd64 package in Ubuntu:
  Confirmed

Bug description:
  This is either a duplicate or return of
  - https://bugs.launchpad.net/ubuntu-cdimage/+bug/1883040
  - https://bugs.launchpad.net/ubuntu-cdimage/+bug/1886148

  Most probably "groovy daily won't boot anymore on some older BIOS
  boxes" or the 1883040 bug, as it's impacting only my older slow boxes
  (but it's also different).

  thumb-drive has been successfully used in QA live & install tests,
  however it won't boot on

  hp dc7700 (c2d-e6320, 5gb, nvidia quadro nvs 290)
  hp dc7900 (c2d-e8400, 4gb, intel 4 series integrated i915)

  The last timing issue also impacted d755-5 (older dell optiplex 755)
  when dc7700 was impacted, but that booted successfully this ISO.

  I see thumb-drive flash as it tries to boot, eventually it stops
  flashing, but no boot. At least 6 attempts made to boot, thumb-drive
  being tested on another 2x d755 boxes..

  (ISO written by mkusb/dus & gnome-disks [#11] is identical)

  ** Expected result

  It boots & is usable

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-cdimage/+bug/1899308/+subscriptions



More information about the foundations-bugs mailing list